The invention represents a software system and method to emulate professional sound effects applied to almost any common instrument, without the use of expensive hardware or significant human effort. The effects are induced in a novel digital form, as an Artificial Neural Network (ANN) by creating completely new embodiments of existing musical effects.
Often, special musical effects are induced into raw audio signals using hardware and/or software, to modify how listeners hear these signals. These effects are achieved through the use of hardware such as effects pedals, and/or effects software. The implementation of these music effects using hardware is expensive, since it is limited to a particular effect or a small number of effects. In addition, effects hardware is cumbersome to use and the effects produced cannot be edited post recording. Existing effects software also tends to be expensive and complex.
UCF researchers have developed a cheaper, more portable and flexible method for audio effects generation, using NeuroEvolution of Augmenting Topologies (NEAT). In this method, audio effects originally implemented by a variety of electronic hardware, can be learned or digitally emulated and then stored in Artificial Neural Networks (ANN). These ANN can then be used to induce the audio effect into raw audio signal files. The audio effects produced by NEAT are an original embodiment of the effects emulated.
- Inexpensive method of inducing musical effects
- Compatible with modern music mixing or composing software
- Eliminates the need for effects hardware as part of the recording process
- Provides flexibility, allowing for post-production of the sound
- Does not violate copyright law, since the digital representation generated is original
- Can be used to produce effects in raw audio signals
- Produce digital musical effects:
- Can be modified post production
- Produced in bundles sold as single units or an entire software package
- Generate audio effects on site
- Develop hardware with the capability of reading and replicating the audio effects stored as an ANN